Artificial intelligence should be viewed as a tool to improve regulatory compliance and not be positioned solely as a risk, insurtech founder Steve Hind says.

Mr Hind, CEO of Sydney start-up Lorikeet, notes concern over AI “hallucinations”, conceding: “We can’t prove it will never [happen]. That’s impossible.”

But he says there is a “deadlock between compliance [departments] and business”, even though organisations may currently breach compliance policies 3% of the time while AI tools may do so at a rate of 0.5%.

“Here is our pro-compliance case for deploying AI ... These products will be seen as a lever to improve compliance posture, not as a compliance risk,” Mr Hind told the sold-out InsurtechLIVE26 in Sydney.

“My take – and what we’re seeing in the market – is actually 2026 will be the year that AI is positioned as a partnership between business and compliance.”

Lorikeet – which raised $54 million last year from investors including the founders of Canva and Australian fintech Airwallex – focuses on building AI agents in complex and well-regulated businesses, including insurance and telehealth.

Mr Hind says many AI agents are “kind of s---house ... Your customers are coming for help and you’re just telling them what they could have read online themselves.”

Lorikeet began with a tool for lost and stolen bank cards. “Shipping out new cards, cancelling cards. Very, very high-risk stuff,” Mr Hind said.

“You’re going to trust an AI agent to go have a conversation with your customers. Is it going to hallucinate? Is it going to promise to pay out a claim? Is it going to tell someone they have a certain damage covered by the policy that isn’t covered?  

“What it has tended to do is pit the business against the compliance department, because the business says, ‘Well, here’s these gains we get from this.’ And then compliance starts asking quite hard questions.”

Mr Hind warns consumers are going to use ChatGPT and other agentic AI products for insurance advice, and while call centres are “tied in knots with what they can and can’t say ... it’s going to take a crack.  

“If your compliance posture is stopping you answering the questions people want in a high-quality way, they will just stop asking you the questions ... If you’re in your 10th compliance meeting of the year about what type of advice you can give while ChatGPT just spews it out, you’re in trouble.  

“The baseline is changing very rapidly. I think it poses a challenge for everyone to say, ‘What will we do with our digital presence that makes it good enough to continue to deserve our customers’ attention?’

“And the businesses that fail to do that, I think, will just stop seeing customers turn up.”